HIV Articles  
CD4 percentage is an independent predictor of survival in patients starting HAART with absolute CD4 cell counts between 200 and 350 cells/μL
  HIV Medicine
Volume 7 Page 383 - September 2006
DM Moore1,3, RS Hogg1,2, B Yip1, K Craib1, E Wood1,3 and JSG Montaner1,3 1BC Centre for Excellence in HIV/AIDS, Vancouver, British Columbia, Canada, 2Department of Health Care and Epidemiology, Faculty of Medicine, University of British Columbia, Vancouver, British Columbia, and 3Department of Medicine, Faculty of Medicine, University of British Columbia, Vancouver, British Columbia, Canada
we analysed the utility of CD4 percentage as a prognostic factor in a subgroup of individuals who would be considered for HAART therapy using current guidelines, but for whom the decision to treat is not yet clearly defined and where absolute CD4 cell counts have been shown not to have predictive value.
To determine the prognostic value of baseline CD4 percentage in terms of patient survival in comparison to absolute CD4 cell counts for HIV-positive patients initiating highly active antiretroviral therapy (HAART).
A population-based cohort study of 1623 antiretroviral therapy-naive HIV-positive individuals who initiated HAART between 1 August 1996 and 30 June 2002 was conducted. Cumulative mortality rates were estimated using Kaplan-Meier methods. Cox proportional hazards regression was used to model the effect of baseline CD4 strata and CD4 percentage strata and other prognostic variables on survival. A subgroup analysis was conducted on 417 AIDS-free subjects with baseline CD4 counts between 200 and 350 cells/μL.
In multivariate models, low CD4 percentages were associated with increased risk of death [CD4%<5, relative hazard (RH)=4.46; CD4% 5-14, RH=2.43; P<0.01 for both] when compared with those subjects with an initial CD4 fraction of 15% or greater, but had less predictive value than absolute CD4 counts. In subgroup analyses where absolute CD4 strata were not associated with mortality, a baseline CD4 fraction below 15% [RH=2.71; 95% confidence interval (CI) 1.20-6.10], poor adherence to therapy and baseline viral load >100 000 HIV-1 RNA copies/mL were associated with an increased risk of death.
CD4 percentages below 15% are independent predictors of mortality in AIDS-free patients starting HAART, including those with CD4 counts between 200 and 350 cells/μL. CD4 percentage should be considered for inclusion in guidelines used to determine when to start therapy.
Between 1 August 1996 and 30 June 2002, a total of 2070 ARV-naive participants aged 18 years and over initiated triple combination therapy consisting of two nucleoside reverse transcriptase inhibitors plus a protease inhibitor (PI) or a nonnucleoside reverse transcriptase inhibitor. Of these, 447 (21.6%) were excluded from this analysis for not having baseline CD4 cell counts, CD4 percentages or viral load measurements available within 6 months prior to the start of ARV therapy. The total study sample was based on the remaining 1623 subjects (78.4%). Participants excluded from the study were slightly younger (median age 37 years vs 38 years; P=0.001), less likely to be male (74.5 vs 83.8%; P=0.001) and more likely to have been prescribed a PI in their initial regimen (69.8 vs 63.7%; P=0.016) than those included in the analysis.
The median baseline CD4 cell count was 230 cells/μL [interquartile range (IQR) 90-380 cells/μL] and the median CD4 percentage was 16% (IQR 9.0-24%). Baseline CD4 count and CD4 percentage were highly correlated (Pearson's r2=0.752; P<0.001). A total of 221 study subjects (13.6%) had clinical AIDS at baseline. A total of 202 nonaccidental deaths were identified in the study population over the follow-up period, with an overall crude mortality rate of 12.5% in a median follow-up time of 45 months (IQR 25-63 months).
Baseline CD4 cell counts were associated with an increased relative hazard (RH) of mortality for all 50 cells/μL strata below 200 cells/μL in univariate analyses in comparison to those subjects with baseline counts >500 cells/μL (Table 1). When strata with similar hazard ratios were combined, three hazard levels were found: baseline CD4 count <50 cells/μL [RH=4.97; 95% confidence interval (CI) 3.51-7.05], 50-199 cells/μL (RH=2.86; 95% CI 2.04-4.00) and >200 cells/μL (RH=1.00). For baseline CD4 percentage, three hazard levels were again found: <5% (RH=3.66; 95% CI 2.51-5.36), 5-14% (RH=2.48; 95% CI 1.81-3.41) and 15% (RH=1.00) (Fig. 1).
When combined in a multivariate model including age, adherence to therapy, initial viral load, and a previous diagnosis of AIDS, a low CD4 percentage remained a highly significant predictor of eventual death [CD4% <5, relative risk (RR)=4.46, 95% CI 2.92-6.79; CD4% 5-14, RR=2.43, 95% CI 1.75-3.38] when compared to those subjects with an initial CD4 fraction of 15% or greater (Table 2). However, when compared to absolute CD4 count groups, CD4 percentage groups had a weaker association with later mortality. In a multivariate model using absolute CD4 count instead of CD4 percentage, the hazards for the low CD4 count strata were higher (absolute CD4 <50 cells/μL, RH=6.07, 95% CI 4.11-8.97; 50-199 cells/μL, RH=2.95, 95% CI 2.08-4.18) relative to counts 200 cells/μL. When both absolute CD4 count groups and CD4 percentage groups were included in the multivariate model, CD4 percentage was not associated with the outcome (data not shown).
The subgroup analysis was conducted on 417 cohort patients (25.7%) who were AIDS-free and started HAART with absolute CD4 counts between 200 and 350 cells/μL (Table 3). Median baseline CD4 count was 280 cells/μL (IQR 240-310) and median CD4 percentage was 18% (IQR 14-24). Absolute CD4 count and CD4 percentage were again correlated (Pearson's r2=0.229, P<0.001), but more weakly than in the whole group. Nonaccidental deaths occurred in 24 patients (crude mortality rate of 5.76%). Baseline CD4 fraction <15% was significantly associated with death in multivariate models (RH=2.71, 95% CI 1.20-6.10). No patients had CD4 fractions below 5% in this group. Poor adherence to therapy (RH=1.18, 95% CI 1.05-1.33) and baseline viral load >100 000 HIV-1 RNA copies/mL (RH=2.57, 95% CI 1.00-6.58) were also associated with an increased risk of mortality.
These analyses demonstrate that baseline CD4 percentages can provide similar prognostic value to absolute CD4 cell counts in treatment-naive individuals who begin HAART. However, in a comparison of the two measures for all patients in this cohort, absolute CD4 cell counts appeared to be a stronger predictor of disease progression. Absolute CD4 counts between 200 and 350 cells/μL were not associated with an increased RH for mortality but, for subjects with baseline CD4 counts within this range and without an AIDS-defining illness, baseline CD4 fractions of less than 15% were significantly associated with increased mortality. These data suggest that CD4 percentage could be added to recommendations regarding when to initiate HAART in asymptomatic HIV-positive individuals with absolute CD4 counts above 200 cells/μL. Further, this analysis has also confirmed the important role of adherence to therapy and viral load measurements >100 000 copies/mL as being important prognostic parameters [13].
This is the first study to examine the utility of baseline CD4 percentage as a prognostic indicator of survival in patients on HAART. A recent study found that baseline CD4 percentage was significantly associated with later risk of developing an AIDS-defining illness, but that this association was weaker than that of absolute CD4 cell count [9]. The authors concluded that CD4 percentage contributed little predictive information beyond that provided by absolute CD4 cell count. Our results agree with this conclusion in the whole group analysis, but show that CD4 percentage is an independent predictor of mortality in the subgroup of AIDS-free patients with absolute CD4 counts between 200 and 350 cells/μL.
Our results also show that absolute CD4 cell counts and CD4 percentages are less correlated at higher absolute counts than they are for subjects with baseline counts <200 cells/μL. As absolute counts are calculated as the product of CD4 percentage and the total lymphocyte count (TLC), this may be a result of there being greater variability in TLCs in subjects with higher CD4 cell counts. Thus, patients with low CD4 percentages but absolute counts >200 cells/μL may, in fact, have spuriously high absolute counts if the TLC is elevated by concurrent infections or other phenomena. It is also possible that other T- or B-cell subsets are present in greater numbers in these patients than in those with lower baseline CD4 cell counts, which resulted in the abnormally low CD4 percentage.
This analysis has a number of limitations. Firstly, nonaccidental mortality may be an imperfect measure of disease progression, as not all deaths will necessarily be HIV-related. However, given the difficulty in clearly identifying what deaths are clearly a result of HIV disease or therapy, this measure seems appropriate. Secondly, the association between disease progression and adherence to therapy could be in the direction of worsening adherence being a consequence of more progressive disease, rather than poor adherence leading to earlier disease progression. However, only adherence in the first year of therapy is used in constructing this variable and the fact that this association is robust even in patients with no symptoms and higher CD4 counts makes this reverse-causation much less likely. Thirdly, this was an observational study in which the patients who presented for therapy, the therapy they received and the time at which they started therapy were not determined by the research team, and so other unmeasured factors may have influenced the outcomes. Finally, the clinical thresholds found here may not represent the exact values at which risk for later mortality increases. Our study lacked the statistical power to delineate thresholds that may exist between the 50 cells/μL increments for absolute CD4 cell count or the 5% increments for CD4 percentage.
In conclusion, previous studies from our centre have shown that absolute CD4 counts between 200 and 350 cells/μL are not predictive of mortality in this cohort, once adherence to therapy is considered [12]. These results suggest that CD4 fractions below 15% are a marker of significant immune compromise in this group and that CD4 percentage should be considered as a recommended parameter to aid clinicians in determining the optimum time to initiate HAART.
Recently revised guidelines from the International AIDS Society-USA (IAS-USA) recommend that highly active antiretroviral therapy (HAART) should be started in all treatment-naive individuals with symptomatic HIV disease, as well as those individuals who are asymptomatic, but who have CD4 counts below 200 cells/μL [1]. For asymptomatic individuals with CD4 counts above 200 cells/μL, the decision of when to start treatment should be individualized and should take into account the rate of decline in CD4 cell count, HIV plasma RNA levels, and the individual risks of toxicity and drug interactions, as well as the patient's interest in antiretroviral (ARV) therapy and their potential to adhere to therapy [1]. Other recommendations have suggested that this threshold should be 350 cells/μL [2], given the evidence that delaying treatment until CD4 counts drop below 200 cells/μL is associated with an increased risk of mortality [3].
The number of CD4 cells expressed as a percentage of the total number of lymphocytes has been used anecdotally by clinicians as another measure to suggest when ARV therapy should be started in asymptomatic individuals with CD4 counts above 200 cells/μL. However, the validity of this measure in terms of its association with survival in patients on HAART has not been evaluated in large prospective studies. Three studies conducted in the pre-HAART era did find that CD4 percentage was a better predictor of disease progression than absolute counts [4-6] and another found that a CD4 fraction of less than 14% was associated with life-threatening opportunistic infections such as Pneumocystis carinii pneumonia (PCP) [7]. In another study, patients with discordant absolute CD4 cell counts and percentages displayed a trend towards increased incidence of PCP and mortality, but these differences were not statistically significant [8]. However, there have been few studies that have compared the relative values of absolute CD4 count and CD4 percentage with respect to their utility in predicting clinically significant outcomes in patients beginning HAART. One recently published study found that CD4 percentage did not provide further value over absolute count in predicting risk of a subsequent AIDS-defining illness in 2185 patients who were followed for a maximum of 6 months after obtaining a discordant absolute CD4 and CD4 percentage laboratory result [9]. However, not all patients (59% in total) in this study received HAART.
As CD4 percentage is commonly included in laboratory reports provided to clinicians of absolute CD4 cell counts at baseline, we designed a study to compare its clinical utility in predicting disease progression with that of absolute CD4 cell count in a cohort of individuals starting HAART in British Columbia, Canada. Furthermore, we analysed the utility of CD4 percentage as a prognostic factor in a subgroup of individuals who would be considered for HAART therapy using current guidelines, but for whom the decision to treat is not yet clearly defined and where absolute CD4 cell counts have been shown not to have predictive value.
The distribution and the population-based monitoring of ARV therapy in British Columbia have been previously described [3]. Briefly, since October 1992, the HIV/AIDS Drug Treatment Program (DTP) of the British Columbia Centre for Excellence in HIV/AIDS (the Centre) has distributed ARV drugs based on guidelines generated by the Therapeutic Guidelines Committee.
Data collection
Physicians enrolling an HIV-infected individual in the DTP must complete a drug request enrolment prescription form, which compiles information on the applicant's address, past HIV-specific drug history, CD4 cell counts, viral load, current drug requests, and enrolling physician data. Typically, persons receiving ARV therapy are monitored by physicians at intervals no longer than 3 months, at which time prescriptions are renewed or modified. At the time of the first refill, participants are asked to provide informed consent for accessing medical electronic records, and complete a participant survey, which elicits information on sociodemographic characteristics, clinical and health status, and alternative therapy use. Both the consent form and the participant survey are optional and a participant's refusal to do either will not limit his or her access to free ARV therapy. The DTP has received ethical approval from the University of British Columbia Ethics Review Committee at its St Paul's Hospital site. The programme also conforms to the Freedom of Information and Protection of Privacy Act of the province.
The Centre recommends that plasma HIV-1 RNA levels and CD4 cell counts be monitored at baseline, at 4 weeks after starting ARV therapy and every 3 months thereafter [10]. Plasma HIV-1 RNA levels were determined using the Roche Amplicor Monitor assay (Roche Diagnostics, Laval, Quebec, Canada) using either the standard method or the ultrasensitive adaptation. CD4 cell counts were measured by flow cytometry, followed by fluorescent monoclonal antibody analysis (Beckman Coulter, Inc., Mississauga, Ontario, Canada).
Study participants
Participants considered in the present study were ARV-naive and were first dispensed triple combination therapy between 1 August 1996 and 30 June 2002. Participants must also have had a CD4 cell count, CD4 percentage and plasma HIV-1 RNA measurement within 6 months of the first ARV start date.
Statistical analyses
The primary endpoint in this analysis was nonaccidental mortality. Deaths occurring during the follow-up period were identified through annual record linkages carried out with the British Columbia Division of Vital Statistics. Cumulative mortality rates were estimated using Kaplan-Meier methods. Event-free subjects were right censored as of 30 June 2003. Cox proportional hazards models were constructed using baseline CD4 strata of 50 cells/μL increments to assess their association with nonaccidental mortality in comparison to study subjects who initiated HAART at baseline cell counts >500 cells/μL. In the same way, Cox models were constructed using baseline CD4 percentages of 5% increments for comparison with study subjects with baseline percentages >40%. Where hazard ratios and confidence intervals were similar between strata, these were aggregated to achieve the final stratifications for analysis in multivariate models along with other baseline factors. Adherence to therapy, defined as the amount of medication that was actually dispensed as a proportion of that needed for the first year of follow-up, was also included in these models. While this is a crude estimate of adherence, previous studies have found it to be highly predictive of clinical outcomes [11,12] Interaction variables were also constructed and included in the multivariate models where significant associations with mortality were found.
The derived CD4 strata were then analysed in the subgroup of patients who did not have an AIDS-defining illness at baseline and initiated therapy with absolute CD4 counts between 200 and 350 cells/μL, a range in which baseline CD4 counts do not have prognostic value [3]. All analyses were performed using SAS software version 8.0 (SAS Institute, Cary, NC, USA).
  icon paper stack View Older Articles   Back to Top